My question here is just an example of a more general problem I'm having of working in local coordinates, so any help would be appreciated.

I am told that $\mathrm{Ric}_{ij}=g^{kl}R_{iklj}$, where $R_{iklj}$ are the components of the curvature tensor $R$. Here $R$ is considered as a $(0,4)$-tensor, related to $R$ considered as a $(1,3)$-tensor by \begin{equation}R(X,Y,Z,W)=g(R(X,Y)Z,W)\end{equation}

I am slightly confused as to how to obtain $\mathrm{Ric}_{ij}=g^{kl}R_{iklj}$, given the definition of Ricci curvature I've been given: \begin{equation} \mathrm{Ric}(v,w)=\mathrm{trace}(x\mapsto R(x,v)w). \end{equation} It seems to me that first I want to convert $R$ from its $(0,4)$ form to its $(1,3)$ form, but with $R=R_{iklj}\sigma^i\otimes\sigma^k\otimes\sigma^l\otimes\sigma^j$ I'm not sure which of the factors I should be converting? Suppose for instance I take the second factor, so \begin{equation} \sigma^k\rightarrow g^{ks} E_s \end{equation} (this is by the usual isomorphism between $TM$ and $T^*M$ on a Riemannian manifold, where $E_i$ is our frame and $\sigma^i$ our coframe). Thus we obtain a $(1,3)$-tensor \begin{equation} \begin{split} R & = g^{ks}R_{iklj}\sigma^i \otimes E_s \otimes \sigma^l \otimes \sigma^j \\ &= {R_i^{~~s}}_{lj}\,\sigma^i \otimes E_s \otimes \sigma^l \otimes \sigma^j \end{split} \end{equation}

Even if this is the correct choice to make, in order to obtain the Ricci tensor (considered as a $(0,2)$-tensor) I must now contract the $E_s$ factor with one of the $\sigma$ factors. Again, I am not sure which factor I should be contracting with to obtain the Ricci tensor. Choosing the $\sigma^l$ factor I obtain

\begin{equation} S = S_{ij}\sigma^i\otimes \sigma^j \end{equation} where $S_{ij} = {R_i^{~~s}}_{sj}$.

Is this tensor $S$ the Ricci tensor? Does $S_{ij}=g^{kl}R_{iklj}$? How am I to know what choices to make (when working in coordinates as above) when changing tensor types/performing contractions, from the definitions I've been given that are not in coordinates? Thanks in advance.


We have that $x \mapsto R(x,v)w$ is a linear map for fixed $v,w$. We would like to find its matrix, so as to be able to extract its trace, which is the Ricci curvature.

Suppose we have a system of local coördinates $x_1, \ldots, x_n$ about a point $p \in M$ of our manifold, and we take the corresponding local basis of the tangent space $\frac{\partial}{\partial x_1}, \ldots, \frac{\partial}{\partial x_n}$. In these coördinates, we have the curvature tensor field $R_{ijkl}$, which we may also regard as a $C^\infty(M)$-multilinear map $\mathfrak X(M)^4 \to \mathcal C^\infty(M)$. Then we try to compute the trace of the map $$ x \mapsto R\left(x,\frac{\partial}{\partial x_i}\right)\frac{\partial}{\partial x_j}. $$ This map, being a linear map between vector spaces, may be written (in the given basis) as a matrix $A \in \mathbb R^{n\times n}$, where we suppose that the linear operator is left-multiplication by $A$. The trace will then be the sum of the diagonal entries of that matrix.

In order to extract the matrix $A$, we note first that we have, denoting the Riemannian metric at $p$ simply by $\langle \cdot, \cdot \rangle$ $$ \left\langle R\left(\frac{\partial}{\partial x_i},\frac{\partial}{\partial x_j}\right)\frac{\partial}{\partial x_k}, \frac{\partial}{\partial x_l} \right\rangle = R_{ijkl} $$ by definition. Noting that the matrix of the form $\langle \cdot, \cdot \rangle$ is given by $$ \left( \left\langle \frac{\partial}{\partial x_i}, \frac{\partial}{\partial x_j} \right\rangle \right)_{i,j} =: G $$ and that the $i,j$-th entry of its inverse is denoted by $g^{ij}$, we get that the $i$-th column of the matrix $A$ is given by $$ \begin{pmatrix} R_{ijk1} & \cdots & R_{ijkn} \end{pmatrix} = \begin{pmatrix} a_{1i} & \cdots & a_{ni} \end{pmatrix}G, $$ and after multiplying by $G^{-1}$ on the right, we get $$ \begin{pmatrix} R_{ijk1} & \cdots & R_{ijkn} \end{pmatrix} G^{-1} = \begin{pmatrix} a_{1i} & \cdots & a_{ni} \end{pmatrix}. $$ But the $m$-th entry of left-hand side equals $$ R_{ijkl} g^{lm} $$ so that we get $\operatorname{Ric}_{jk} = R_{ijkl} g^{li}$.

  • $\begingroup$ @kt77 thanks for giving me no points for the answer. $\endgroup$ – AlgebraicsAnonymous Jan 7 '18 at 20:08

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.